Running Performance Max for a Hotel Without Letting Google Burn the Budget

Performance Max is the campaign type Google would very much like every hotel to run, and I get why. You hand over your assets, your feed, and your budget, and the machine spreads it across Search, Display, YouTube, Gmail, Maps, and Discover without you having to think too hard. For a lot of independent hoteliers, “without having to think too hard” is exactly the trap.

Here is the thing I tell every owner before we turn one on: PMax is a black box that is structurally incentivized to find the cheapest conversions it can, and the cheapest conversions in your whole funnel are the people already typing your hotel’s name into Google. Those guests were going to book you anyway. If you are not careful, you end up paying Google a commission to recapture demand you already created. It feels like the campaign is working. The dashboard is green. Meanwhile your “incremental” bookings are mostly people who walked in the front door you’d left wide open.

So this is how I actually run Performance Max for a hotel without letting it quietly burn the budget on your own brand.

First, decide whether you even need it

PMax is a MOFU/BOFU tool. It is not a substitute for being findable in the first place. If your hotel is invisible in organic and on Maps, paid media is just renting visibility you should own. I’d rather fix the foundation first with hotel SEO and a clean Google Business Profile, then layer paid on top to capture incremental demand. Paid is the accelerant, not the engine.

A quick gut check before spending a dollar:

• Do you already rank on page one for your own hotel name, or are the OTAs outranking you for it? Fix that with organic first, it’s cheaper.
• Is your direct booking path fast and frictionless? If the booking engine is clunky, paid traffic just leaks out the bottom.
• Can you measure a real direct booking as a conversion with revenue? If not, stop, because you will be flying blind.

If those are roughly in place, PMax can earn its keep. If they’re not, you’re about to pay to fill a leaky bucket.

Structure: one campaign, themed asset groups, ruthless intent

The single most common mistake I see is a hotel running one PMax campaign with one giant asset group called “Hotel” and every audience signal Google suggested. That’s how you get blended garbage you can’t read.

I structure around how people actually shop a stay. A simple, defensible setup:

Each asset group gets its own headlines, descriptions, images, logo, and crucially its own landing page. I do not send PMax traffic to the homepage. I send it to a page with live rates, real photos, and a visible book-direct reason to believe, the kind of conversion-focused page I’d build in a book-direct CRO engagement. The homepage is a hub; a paid click needs a destination that closes.

The feed is your steering wheel

If you run an ecommerce PMax, the product feed is everything. Hotels don’t have a Merchant Center product feed in the same way, but you absolutely have feed-shaped signals, and they are the most underused steering wheel in the whole setup.

A few that matter:

• Page feed / URL list. Give PMax a curated list of your best landing pages rather than letting it crawl your whole site and decide for itself. This is how you keep it on rooms and offers pages and off your blog, careers page, and contact form.
• Business and location data. Connect your Google Business Profile location so PMax has accurate location, photos, and hours to draw on. A well-maintained GBP, which I cover in the GBP playbook, does double duty here.
• Hotel rates and availability. If you participate in Google’s hotel ads ecosystem and metasearch, that live-rate signal is gold. I get into the mechanics of that in the metasearch piece, but the short version: accurate, live direct rates feeding the system make your paid placements far more relevant.

Garbage in, garbage out applies brutally here. If your feed points PMax at a slow homepage with stale photos and no live rate, no amount of bid magic saves you.

Audience signals are hints, not handcuffs

People misunderstand audience signals in PMax. They are not targeting in the old sense. They are a starting nudge that tells Google’s model where to begin looking before it goes off and finds its own pockets of demand.

What I feed it:

• Your first-party data. Past direct bookers, your email list, high-value repeat guests. This is the most valuable signal you have, and most independents never upload it.
• Custom segments built from search terms like “boutique hotel [city],” competitor hotel names, and high-intent stay phrases.
• Your converting website visitors, the people who hit the booking engine but didn’t complete.

Then I let the model expand from there, but I watch where it actually goes. Which brings us to the part nobody enjoys but everybody needs.

Exclusions: this is where you stop the bleed

Here’s the heart of it. Left to its own devices, PMax will serve on your brand terms and claim the credit. So you build fences.

Brand exclusions. Google now lets you apply brand lists to PMax to keep it off your own hotel name and common misspellings. Turn this on. If someone searches your exact property name, I want that handled by a tightly controlled brand search campaign (or won organically for free), not absorbed into PMax where it inflates the numbers and hides the truth. Recapturing your own brand is exactly the dynamic I describe in how OTAs steal your search, except here you’re paying Google to do it to yourself.

Negative keywords. You historically had to ask your Google rep to apply account-level negatives to PMax; now there’s more self-serve control. Either way, get negatives in for: your brand name, “[hotel] reviews,” “[hotel] jobs,” “cancel reservation,” and anything that screams existing-guest or non-booker intent.

Placement and content exclusions. Account-level placement exclusions and content suitability settings keep your spend off junk apps, made-for-advertising sites, and inappropriate content. For a boutique brand, where you show up is part of the brand.

URL expansion: turn it off, or tightly control it. Final URL expansion lets PMax send traffic to any page on your site it thinks will convert. Sounds helpful. In practice it’ll route paid clicks to your blog or a random amenities page. I switch it off and feed an explicit page list, or I leave it on only with a long exclusion list. You bought that click; you decide where it lands.

The whole game with Performance Max is making the machine optimize toward bookings you would not have gotten anyway. Every exclusion you add is you telling Google: stop taking credit for demand I already created. That single reframe is worth more than any bidding trick.

Measuring incrementality, not vanity conversions

The in-platform conversion number is the most flattering and least trustworthy figure in the whole account. PMax is very good at showing you a great ROAS that is really just your existing brand demand wearing a paid costume.

So I judge it differently:

• Watch total direct revenue, not campaign-attributed revenue. Did direct bookings actually go up after launch, or did the campaign just reassign credit?
• Run a geo or time-based holdout if budget allows. Pause PMax in a region or for a window and see whether direct bookings actually fall. If they don’t move, you weren’t buying incremental anything.
• Mine the search-terms and “search categories” insights weekly. This is where you catch PMax drifting back onto brand terms or junk queries, and where your next round of negatives comes from.
• Tie it back to the real economics. A direct booking that dodges a 15 to 25 percent OTA commission is worth a lot more than the same booking through an OTA. That margin is the whole reason we’re doing this, and I run the book-direct math on every account so we know what an incremental direct booking is actually worth before we set targets.

A realistic launch sequence

If I’m standing up a hotel’s Performance Max from scratch, the order looks like this:

1. Fix tracking first. Real booking-engine conversions with revenue values, server-side if possible. No tracking, no campaign.
2. Set up brand exclusions and the negative list before launch, not after you’ve already paid to recapture your name for a month.
3. Launch one campaign, two or three themed asset groups, each with a dedicated landing page and full asset coverage (Google rewards complete asset groups).
4. Final URL expansion off, page feed on, first-party audience signals uploaded.
5. Leave it alone for the learning window. Roughly two to six weeks of noise. Resist the urge to tweak daily; you’ll just reset learning and never get clean data.
6. Then read the search-terms insights, add negatives, and decide on incrementality. Scale the asset groups that produce real lift, throttle the ones that don’t.

No part of this guarantees a ranking or a fixed number of bookings, and anyone promising you that is selling something. What it does is tilt the odds: it makes the budget chase genuinely new demand instead of quietly buying back guests you’d already earned, so more of your bookings come direct and more of that margin stays with you instead of leaking to commissions.

Performance Max is not the enemy and it’s not magic. It’s a powerful, slightly greedy tool that does exactly what you constrain it to do. Constrain it well and it’s one of the better ways to claw back direct bookings. Leave it wide open and it’ll cheerfully spend your money proving how well it captures demand you created for free.

Want a second set of eyes before you flip it on?

If you’re about to launch Performance Max, or you’ve got one running and you suspect it’s just eating your brand search, that’s exactly the kind of thing I dig into on a free intro call. Bring your search-terms report and we’ll figure out what’s actually incremental. Or if you’d rather start with the structural fixes that make paid worth running at all, take a look at book-direct CRO and the AI visibility work that’s increasingly where boutique demand begins.